Immune inflammation indicators and ALBI score to predict liver cancer in HCV- patients treated with direct-acting antivirals

Background: Unexpectedly high occurrence or recurrence rate of hepatocellular carcinoma (HCC) has been observed in patients with chronic hepatitis C receiving direct-acting antivirals (DAAs) therapy. Aims: We evaluated the predictive value of albumin-bilirubin (ALBI) score and immune-inflammation indicators to identify the risk of occurrence or recurrence of HCC in patients treated with DAAs in a real life setting. Methods: In this retrospective cohort study, we analysed data from 514 patients with cirrhosis who were prospectively enrolled for treatment with DAAs. We assessed baseline neutrophil to lymphocyte ratio (NLR), systemic immune-inflammation index (SII), platelet to lymphocyte ratio (PLR), aspartate aminotransferase-lymphocyte ratio (ALRI) index and ALBI score. Results: In patients with no history of HCC (N=416), increased AST, bilirubin, ALRI, and ALBI score, and decreased albumin and platelets were significantly associated with an increased risk of HCC development, at univariate analysis. At multivariate analysis, increase in ALBI grade (p = 0.038, HR: 2.35, 95% CI: 1.05–5.25) and decrease in AC CE PT ED M AN US CR IP T


Introduction
Interferon(IFN)-free regimen using new direct acting antivirals (DAAs) has represented a turning point in the treatment of patients with chronic hepatitis C [1].
The impact of DAAs-based treatment on the development of hepatocellular carcinoma (HCC) in patients with cirrhosis has grown controversial due to potential clinical implications, particularly for HCC recurrence after a successful curative treatment.
The mechanism that could explain such a high rate of tumor recurrence after DAAs treatment is one of the main topics of these studies. Microenvironment and viralinduced inflammation are supposed to play a key role in chronic liver injury and tumor initiation [10]. However, the human immune system itself has also an anti-tumor function [11]. Overall, there is a complex yet fragile balance between the pro-and antitumor functions of the immune system. Some studies have proposed that DAAs treatment could modify the natural killer function and the expression of IFN response genes [12,13].

A C C E P T E D M A N U S C R I P T
The albumin-bilirubin (ALBI) score was first proposed by Johnson et al. [24], with the aim to obviate the need of subjective variables, such as ascites and encephalopathy, in the Child-Pugh grading system, for liver function evaluation in HCC patients.
The objective of this study was to evaluate the predictive value of ALBI score and immune-inflammation indicators to identify the risk of HCC occurrence or recurrence in patients treated with DAAs for chronic hepatitis C.

2.1Patient selection and treatment
In this retrospective cohort study, we analysed data from all the consecutive HCVinfected cirrhotic patients treated with DAAs in seven hepatology centers in the area of Bologna, Italy, between January 2015 and August 2016. Follow-up ended in January 2017. Data were first retrieved from the electronic regional registry database (Piattaforma SOLE). All the additional data were obtained from the individual patient records.
Eligibility for treating hepatitis C with DAAs was assessed for each patient following the priority criteria established by the national registry of the Italian Medicines Agency (AIFA). When possible, alternative treatment options were sought at the clinician's discretion.
Diagnosis of liver cirrhosis was established if at least one of the following was present: 1) previous liver biopsy with stage 4 fibrosis with METAVIR score; 2) presence of esophageal and/or gastric varices at endoscopy; 3) liver stiffness higher than 12 kPa at transient elastography by FibroScan (Echo Sense, Paris, France) [25,26].
Patients with previous liver transplantation or with treated HCC without radiological complete response before starting DAAs were excluded.
The database included 688 patients treated with different regimens of DAAs (105 with and 563 without HCC history). Since our analysis was limited to patients with liver cirrhosis, 144 were excluded because of a F3 METAVIR score or a transient elastography result of less than 12 kPa. Further 10 patients (5 in each category) were not included in the analysis due to liver transplantation (Fig. 1).
Before starting antiviral therapy, all patients with no HCC history underwent abdomen ultrasound (US). If a potential focal lesion was detected in the liver, the diagnostic work-up was completed with contrast-enhanced ultrasonography (CEUS), and a subsequent computerized tomography (CT) scan or magnetic resonance imaging (MRI) was performed to exclude the presence of HCC. All the patients with a history of HCC underwent ultrasound and CT scan or MRI to exclude recurrent HCC.
All patients were followed-up after EOT. During follow-up, patients repeated US evaluation as recommended by the surveillance programme guidelines [27]; in case of suspicion of HCC development, CEUS/CT/MRI were carried out to determine the presence of HCC.
Virological response to therapy was assessed by quantitative HCV-RNA determination, using real-time PCR with a limit of detection of 15 IU/ml.
All blood values were obtained at baseline and at the end of the treatment.

Statistical Analysis
Data were summarized by mean ± standard deviation or median and minimum or maximum value for continuous variables, and by frequency and percentage for With regard to the time to HCC detection (occurrence or recurrence), we performed distinct analyses for patients with or without a history of HCC.
The Nelson-Aalen estimator was used to obtain cumulative hazard rates at specific follow-up time points within the two groups of patients (with and without a history of HCC). Kaplan-Meier curves were used to represent the time to HCC onset since the beginning of DAA treatment. Time was censored at the time of death, last patient contact or end of follow-up. Cox regression was used to evaluate the association between inflammation indices and other covariates and the time to HCC. When not specified, continuous covariates were included in the models as they were, that is, without dichotomization or categorization. For ease of interpretation, some of the reported hazard ratios (HRs) for continuous covariates correspond to a 10-unit increase in the corresponding variable rather than a 1-unit increase.
The underlying assumption of proportional hazard was tested on the basis of Schoenfeld residuals.
Given the limited number of HCC events and thus the need for a more parsimonious model as well as the presence of correlation among variables, to specify the final Cox regression model we decided to focus our attention of the inflammation indices rather than their single components. Due to the presence of missing values for the continuous variables, multiple imputation using multivariate normal distribution with ten imputed datasets was performed [28].
NLR was computed as the ratio of the absolute neutrophil count to the absolute lymphocyte count, PLR as the ratio of absolute platelet count to the absolute lymphocyte count, SII as platelet count × neutrophil count/lymphocyte count, and ALRI as the ratio of aminotransferase to absolute lymphocyte count. ALBI was obtained through the following formula: (log10 bilirubin × 0.66) + (albumin × -0.085) where bilirubin is in µmol/L and albumin in g/L. PALBI was obtained using the following formula: (2.02 × log10 bilirubin) + [-0.37 × (log10 bilirubin) 2 ] + (-0.04 × albumin) + (-3.48 × log10 platelets) + [1.01 × (log10 platelets) 2 ]. For patients with a history of HCC, the time from HCC treatment until start of DAA treatment was computed and categorized as follows: < 6 months, 6 -12 months, 12 -24 months, and ≥ 24 months.
Exploratory analyses were performed to test for differences in the mean values of clinical parameters at three different time points (baseline, end-of-treatment, and at 12 weeks of follow-up). An ANalysis Of VAriance (ANOVA) for repeated measurements was used for this purpose with Bonferroni adjustment for post-hoc comparisons. The analysis was performed only on patients with data at all three time points.
To evaluate the effect of changes in specific clinical parameters from the baseline to the end-of-treatment, Cox regression models were fitted including the absolute change (i.e. lymphocytes at the end-of-treatment minus lymphocytes at baseline) as covariate of interest and adjusting for the value of the clinical parameter at the baseline. This exploratory analysis was performed only on patients with data at both time points.
A p-value less than 0.05 was considered statistically significant. Analyses were performed using STATA 14.0 (College Station, Texas, USA) and R version 3.4.0 statistical software (package "rms" for drawing the nomogram).

Patient characteristics
We performed our analysis on the 514 consecutive patients with HCV-related liver cirrhosis treated with different DAA regimens between January 2015 and August 2016.
We split the population into two groups for analysis: patients with history of HCC (98) and without (416).
The median follow-up was 18.04 months (range 0.43-26.41) from the start of the treatment. The baseline characteristics of the two groups before starting DAA therapy are shown in Table 1. HCC was detected and confirmed by at least two independent imaging techniques, or biopsy, in 30 out of 98 patients (30.6%) with a history of HCC, and in 29 out of 416 patients (7.0%) without a history of HCC.
At univariable analysis, applying the cut-off values published by Johnson et al [24] to ALBI grade, patients with ALBI grade 2 or 3 had a risk of HCC occurrence over three times higher than those with ALBI grade 1 (p = 0.01, HR: 3.22, 95% CI: 1.32-7.86).
Using a cut-off value of 100 x 10 9 /L for platelets, patients with platelets < 100 x10 9 /L had about a two-fold higher risk of HCC occurrence than patients with platelets > 100

Inflammatory parameters changes after treatment
At baseline, patients without a history of HCC had a higher number of neutrophils and lymphocytes than those with a history of HCC (neutrophils: 2.96 ± 1.33 vs 2.59 ± 1.10 x10 9 /L respectively; lymphocytes: 1.70 ± 0.85 vs 1.37 ± 0.77 x10 9 /L, respectively). As shown in Supplementary Table 1, baseline NLR, PLR and SII significantly increased and ALRI significantly decreased at the end of DAA treatment in patients without a history of HCC. Conversely, no changes were observed in patients with a history of HCC apart from ALRI that resulted significantly decreased.

Discussion
This study evaluated ALBI score and immune-inflammation parameters as potential predictors of HCC occurrence or recurrence in patients treated with DAA for chronic hepatitis C.
Some recent studies have shown unexpected high HCC recurrence rates after DAA therapy [2][3][4], but other studies contradicted these findings [5][6][7][8][9]. Given these contrasting data, it is pivotal to identify the patients at a higher risk and to determine the possible mechanisms associated with early occurrence or recurrence.

A C C E P T E D M A N U S C R I P T
Our study considered a homogenous population of cirrhotic patients divided into two subgroups characterized by the presence of absence of a history of HCC. We found that in patients without a history of HCC, the risk of HCC development seemed dependent on the stage of liver disease (as evidenced by the ALBI score and platelet count). Liver function impairment and cirrhosis remain the main predictors of the risk of HCC for these patients as well as for the untreated patients with hepatitis C [41].
Interestingly, our data show that pre-treatment liver stiffness was not associated with the prediction of the risk to develop HCC at univariate analysis, probably suggesting that liver dysfunction is a better predictor than the stage of liver fibrosis.
Platelets are involved in thrombosis, inflammatory responses, liver regeneration [42][43][44], and the regulation of angiogenesis [45]. Low platelet count in cirrhosis is associated with a higher risk of developing HCC representing an indirect marker of advanced disease [46].
A nomogram was developed to predict the risk of HCC recurrence in this population at 1 year. This graphical representation may help computing the patient's risk of HCC within a specific timeframe and potentially planning differentiated follow-up schedules.
For example, a patient without previous HCC with an ALBI score of -3 (ALBI grade 1 category) and a platelet count of 150 x10 9 /L would have a total of 95.5 points (ALBI grade = 12.5 points and platelets = 83 points). For a patient with these characteristics, the predicted probability of HCC recurrence in the first year of start of DAA therapy is 0%. Conversely, for a patient with ALBI score of -1.40 (ALBI grade 2 category) and a platelet count of 50 x10 9 /L would have a total of 127 points (ALBI grade = 32 points and platelets = 95 points) with a risk of recurrence HCC of about 20%. The nomogram can be used to evaluate the HCC risk for each patient and plan a closer follow-up in high-risk patients.
Hepatic function does not seem to predict the risk of cancer recurrence in patients with a history of HCC, although ALRI index seems to be a good predictor. One explanation could be that a dysregulation of the antitumor response after the sharp decrease in HCV viral load induced by DAA therapy may lead to tumor recurrence. Recurrence could be accelerated by DAA effects, boosting the growth of undetectable HCC as a consequence of a perturbation of the immune surveillance, caused by a swift clearance of HCV.
ALRI index is an inflammation and immune-based prognostic score. Several studies have shown that ALRI index can predict therapy outcome in patients with HCC [14][15][16].
ALRI index can reflect the efficacy of the immune response based on lymphocytes and hepatic index aspartate aminotransferase.
We reported different results in two different populations (patients without a history of HCC and patients with a history of HCC). The greater risk factor for the development of HCC in patients without a history of HCC is the severity of liver cirrhosis, similarly to the tpopulation no treated with DAA. Conversely, we believe that the immuno-related factor was the only predictor of occurrence in patients with a history of HCC because they may had already carried an immunologic switch that had promoted cancer development.
Other interesting results of this study are the significant decrease in lymphocyte counts and the significant increase in neutrophils during DAA therapy. We think that this created a favorable microenvironment for the growth of a misinterpreted focus of cancer cells that had promoted HCC progression. Other markers tested in this study (NLR, PLR and SII) showed an increase during treatment with DAA. This imbalance reflects an increase in the inflammatory state during treatment, allowing us to conclude that this change may underlie HCC occurrence and recurrence during DAA therapy.
With this respect, a recent study by Villani et al. [47] demonstrated that DAA therapy induces a rapid reduction in TNF-α, IL-10, and induces an increase in serum VEGF. In particular, IL-10 is a immunoregulatory molecule involved in anti-inflammatory processes in chronic HCV infection. During persistent infections the virus exploits the production of IL-10 by dentritic cells (DCs) to exhaust antiviral T cells. High IL-10 levels produced by DCs suppress their antigen presenting capacity and lead to inefficient T cell activation. Chronic antigen presence further exhausts T cells and induce IL-10 production. T cells therefore become tolerant to viral antigens and infection persists [48]. Moreover, IL-10 has pleiotropic and controversial functions in tumor biology, as it seems to function at the crossroads between immune stimulation and immune suppression in cancer [49]. For this reason evaluation of ALRI index allows to evaluate the patient's general inflammatory state and to predict which patients are at greater risk of relapse after DAA treatment. This hypothesis, however, needs to be supported in a specific translational trial.
Among the limitations of this study is its retrospective nature, which might have precluded the collection of data on potential confounding factors or important prognostic factors. However, all cases were consecutively selected, thus reducing any potential bias. Missing data were imputed using multiple imputation, a method proven superior to single value imputation methods (i.e. mean imputation), and the complete case approach, which generally generates problems in case of missing at random data, with the effect of reducing the sample size, and in case of missing not at random data, with the effect of producing biased estimates [28]. Finally, another limitation is the relatively short follow-up of patients.
In conclusion, to our knowledge, this is the first study to have investigated the risk factors of HCC development after DAA treatment using immune inflammation indicators and ALBI score. Our results indicate that an impaired liver function, but not immune inflammation indicators, is an independent risk factor for the occurrence of HCC after DAA in patients without a history of HCC. On the contrary, the immuneinflammation ALRI index seems to be an independent predictor of recurrence in patients with a history of HCC. Furthermore, we demonstrated that DAA therapy leads to an increase in neutrophils and a decrease in lymphocytes, resulting in a potential imbalance and a subsequent favorable microenvironment for the growth of cancer cells.
Low cost, easy determination, and reproducibility of these markers could be usefully in clinical practice. Further studies in larger cohorts of patients are needed to confirm these results. Collection and assembly of data: All authors